includes/upload/UploadBase.php

   1 <?php
   2 /**
   3  * @file
   4  * @ingroup upload
   5  *
   6  * UploadBase and subclasses are the backend of MediaWiki's file uploads.
   7  * The frontends are formed by ApiUpload and SpecialUpload.
   8  *
   9  * See also includes/docs/upload.txt
  10  *
  11  * @author Brion Vibber
  12  * @author Bryan Tong Minh
  13  * @author Michael Dale
  14  */
  15
  16 abstract class UploadBase {
  17         protected $mTempPath;
  18         protected $mDesiredDestName, $mDestName, $mRemoveTempFile, $mSourceType;
  19         protected $mTitle = false, $mTitleError = 0;
  20         protected $mFilteredName, $mFinalExtension;
  21         protected $mLocalFile;
  22
  23         const SUCCESS = 0;
  24         const OK = 0;
  25         const EMPTY_FILE = 3;
  26         const MIN_LENGTH_PARTNAME = 4;
  27         const ILLEGAL_FILENAME = 5;
  28         const OVERWRITE_EXISTING_FILE = 7;
  29         const FILETYPE_MISSING = 8;
  30         const FILETYPE_BADTYPE = 9;
  31         const VERIFICATION_ERROR = 10;
  32
  33         # HOOK_ABORTED is the new name of UPLOAD_VERIFICATION_ERROR
  34         const UPLOAD_VERIFICATION_ERROR = 11;
  35         const HOOK_ABORTED = 11;
  36         const FILE_TOO_LARGE = 12;
  37
  38         const SESSION_VERSION = 2;
  39         const SESSION_KEYNAME = 'wsUploadData';
  40
  41         static public function getSessionKeyname() {
  42                 return self::SESSION_KEYNAME;
  43         }
  44
  45         public function getVerificationErrorCode( $error ) {
  46                 $code_to_status = array(self::EMPTY_FILE => 'empty-file',
  47                                                                 self::FILE_TOO_LARGE => 'file-too-large',
  48                                                                 self::FILETYPE_MISSING => 'filetype-missing',
  49                                                                 self::FILETYPE_BADTYPE => 'filetype-banned',
  50                                                                 self::MIN_LENGTH_PARTNAME => 'filename-tooshort',
  51                                                                 self::ILLEGAL_FILENAME => 'illegal-filename',
  52                                                                 self::OVERWRITE_EXISTING_FILE => 'overwrite',
  53                                                                 self::VERIFICATION_ERROR => 'verification-error',
  54                                                                 self::HOOK_ABORTED =>  'hookaborted',
  55                 );
  56                 if( isset( $code_to_status[$error] ) ) {
  57                         return $code_to_status[$error];
  58                 }
  59
  60                 return 'unknown-error';
  61         }
  62
  63         /**
  64          * Returns true if uploads are enabled.
  65          * Can be override by subclasses.
  66          */
  67         public static function isEnabled() {
  68                 global $wgEnableUploads;
  69                 if ( !$wgEnableUploads ) {
  70                         return false;
  71                 }
  72
  73                 # Check php's file_uploads setting
  74                 if( !wfIniGetBool( 'file_uploads' ) ) {
  75                         return false;
  76                 }
  77                 return true;
  78         }
  79
  80         /**
  81          * Returns an array of permissions that is required to upload a file
  82          *
  83          * @return array
  84          */
  85         public static function getRequiredPermissions() {
  86                 return array( 'upload', 'edit' );
  87         }
  88         /**
  89          * Returns true if the user can use this upload module or else a string
  90          * identifying the missing permission.
  91          * Can be overriden by subclasses.
  92          */
  93         public static function isAllowed( $user ) {
  94                 foreach ( self::getRequiredPermissions() as $permission ) {
  95                         if ( !$user->isAllowed( $permission ) ) {
  96                                 return $permission;
  97                         }
  98                 }
  99                 return true;
 100         }
 101
 102         // Upload handlers. Should probably just be a global.
 103         static $uploadHandlers = array( 'Stash', 'File', 'Url' );
 104
 105         /**
 106          * Create a form of UploadBase depending on wpSourceType and initializes it
 107          */
 108         public static function createFromRequest( &$request, $type = null ) {
 109                 $type = $type ? $type : $request->getVal( 'wpSourceType', 'File' );
 110
 111                 if( !$type ) {
 112                         return null;
 113                 }
 114
 115                 // Get the upload class
 116                 $type = ucfirst( $type );
 117
 118                 // Give hooks the chance to handle this request
 119                 $className = null;
 120                 wfRunHooks( 'UploadCreateFromRequest', array( $type, &$className ) );
 121                 if ( is_null( $className ) ) {
 122                         $className = 'UploadFrom' . $type;
 123                         wfDebug( __METHOD__ . ": class name: $className\n" );
 124                         if( !in_array( $type, self::$uploadHandlers ) ) {
 125                                 return null;
 126                         }
 127                 }
 128
 129                 // Check whether this upload class is enabled
 130                 if( !call_user_func( array( $className, 'isEnabled' ) ) ) {
 131                         return null;
 132                 }
 133
 134                 // Check whether the request is valid
 135                 if( !call_user_func( array( $className, 'isValidRequest' ), $request ) ) {
 136                         return null;
 137                 }
 138
 139                 $handler = new $className;
 140
 141                 $handler->initializeFromRequest( $request );
 142                 return $handler;
 143         }
 144
 145         /**
 146          * Check whether a request if valid for this handler
 147          */
 148         public static function isValidRequest( $request ) {
 149                 return false;
 150         }
 151
 152         public function __construct() {}
 153
 154         /**
 155          * Initialize the path information
 156          * @param $name string the desired destination name
 157          * @param $tempPath string the temporary path
 158          * @param $fileSize int the file size
 159          * @param $removeTempFile bool (false) remove the temporary file?
 160          * @return null
 161          */
 162         public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
 163                 $this->mDesiredDestName = $name;
 164                 $this->mTempPath = $tempPath;
 165                 $this->mFileSize = $fileSize;
 166                 $this->mRemoveTempFile = $removeTempFile;
 167         }
 168
 169         /**
 170          * Initialize from a WebRequest. Override this in a subclass.
 171          */
 172         public abstract function initializeFromRequest( &$request );
 173
 174         /**
 175          * Fetch the file. Usually a no-op
 176          */
 177         public function fetchFile() {
 178                 return Status::newGood();
 179         }
 180
 181         /**
 182          * Return true if the file is empty
 183          * @return bool
 184          */
 185         public function isEmptyFile() {
 186                 return empty( $this->mFileSize );
 187         }
 188
 189         /**
 190          * Return the file size
 191          * @return integer
 192          */
 193         public function getFileSize() {
 194                 return $this->mFileSize;
 195         }
 196
 197         /**
 198          * Append a file to the Repo file
 199          *
 200          * @param $srcPath String: path to source file
 201          * @param $toAppendPath String: path to the Repo file that will be appended to.
 202          * @return Status Status
 203          */
 204         protected function appendToUploadFile( $srcPath, $toAppendPath ) {
 205                 $repo = RepoGroup::singleton()->getLocalRepo();
 206                 $status = $repo->append( $srcPath, $toAppendPath );
 207                 return $status;
 208         }
 209
 210         /**
 211          * @param $srcPath String: the source path
 212          * @return the real path if it was a virtual URL
 213          */
 214         function getRealPath( $srcPath ) {
 215                 $repo = RepoGroup::singleton()->getLocalRepo();
 216                 if ( $repo->isVirtualUrl( $srcPath ) ) {
 217                         return $repo->resolveVirtualUrl( $srcPath );
 218                 }
 219                 return $srcPath;
 220         }
 221
 222         /**
 223          * Verify whether the upload is sane.
 224          * @return mixed self::OK or else an array with error information
 225          */
 226         public function verifyUpload( ) {
 227                 /**
 228                  * If there was no filename or a zero size given, give up quick.
 229                  */
 230                 if( $this->isEmptyFile() ) {
 231                         return array( 'status' => self::EMPTY_FILE );
 232                 }
 233
 234                 /**
 235                  * Honor $wgMaxUploadSize
 236                  */
 237                 global $wgMaxUploadSize;
 238                 if( $this->mFileSize > $wgMaxUploadSize ) {
 239                         return array( 'status' => self::FILE_TOO_LARGE );
 240                 }
 241
 242                 /**
 243                  * Look at the contents of the file; if we can recognize the
 244                  * type but it's corrupt or data of the wrong type, we should
 245                  * probably not accept it.
 246                  */
 247                 $verification = $this->verifyFile();
 248                 if( $verification !== true ) {
 249                         return array(
 250                                 'status' => self::VERIFICATION_ERROR,
 251                                 'details' => $verification
 252                         );
 253                 }
 254
 255                 /**
 256                  * Make sure this file can be created
 257                  */
 258                 $result = $this->validateNameAndOverwrite();
 259                 if( $result !== true ) {
 260                         return $result;
 261                 }
 262
 263                 $error = '';
 264                 if( !wfRunHooks( 'UploadVerification',
 265                                 array( $this->mDestName, $this->mTempPath, &$error ) ) ) {
 266                         // @fixme This status needs another name...
 267                         return array( 'status' => self::HOOK_ABORTED, 'error' => $error );
 268                 }
 269
 270                 return array( 'status' => self::OK );
 271         }
 272
 273         /**
 274          * Verify that the name is valid and, if necessary, that we can overwrite
 275          *
 276          * @return mixed true if valid, otherwise and array with 'status'
 277          * and other keys
 278          **/
 279         public function validateNameAndOverwrite() {
 280                 $nt = $this->getTitle();
 281                 if( is_null( $nt ) ) {
 282                         $result = array( 'status' => $this->mTitleError );
 283                         if( $this->mTitleError == self::ILLEGAL_FILENAME ) {
 284                                 $result['filtered'] = $this->mFilteredName;
 285                         }
 286                         if ( $this->mTitleError == self::FILETYPE_BADTYPE ) {
 287                                 $result['finalExt'] = $this->mFinalExtension;
 288                         }
 289                         return $result;
 290                 }
 291                 $this->mDestName = $this->getLocalFile()->getName();
 292
 293                 /**
 294                  * In some cases we may forbid overwriting of existing files.
 295                  */
 296                 $overwrite = $this->checkOverwrite();
 297                 if( $overwrite !== true ) {
 298                         return array(
 299                                 'status' => self::OVERWRITE_EXISTING_FILE,
 300                                 'overwrite' => $overwrite
 301                         );
 302                 }
 303                 return true;
 304         }
 305
 306         /**
 307          * Verify the mime type
 308          * @param $magic MagicMime object
 309          * @param $mime string representing the mime
 310          * @return mixed true if the file is verified, an array otherwise
 311          */
 312         protected function verifyMimeType( $magic, $mime ) {
 313                 global $wgVerifyMimeType;
 314                 if ( $wgVerifyMimeType ) {
 315                         wfDebug ( "\n\nmime: <$mime> extension: <{$this->mFinalExtension}>\n\n");
 316                         if ( !$this->verifyExtension( $mime, $this->mFinalExtension ) ) {
 317                                 return array( 'filetype-mime-mismatch' );
 318                         }
 319
 320                         global $wgMimeTypeBlacklist;
 321                         if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
 322                                 return array( 'filetype-badmime', $mime );
 323                         }
 324
 325                         # Check IE type
 326                         $fp = fopen( $this->mTempPath, 'rb' );
 327                         $chunk = fread( $fp, 256 );
 328                         fclose( $fp );
 329                         $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
 330                         $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
 331                         foreach ( $ieTypes as $ieType ) {
 332                                 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
 333                                         return array( 'filetype-bad-ie-mime', $ieType );
 334                                 }
 335                         }
 336                 }
 337
 338                 return true;
 339         }
 340
 341         /**
 342          * Verifies that it's ok to include the uploaded file
 343          *
 344          * @return mixed true of the file is verified, array otherwise.
 345          */
 346         protected function verifyFile() {
 347                 $this->mFileProps = File::getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 348                 $this->checkMacBinary();
 349
 350                 # magically determine mime type
 351                 $magic = MimeMagic::singleton();
 352                 $mime = $magic->guessMimeType( $this->mTempPath, false );
 353
 354                 # check mime type, if desired
 355                 $status = $this->verifyMimeType( $magic, $mime );
 356                 if ( $status !== true ) {
 357                         return $status;
 358                 }
 359
 360                 # check for htmlish code and javascript
 361                 if( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
 362                         return array( 'uploadscripted' );
 363                 }
 364                 if( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 365                         if( $this->detectScriptInSvg( $this->mTempPath ) ) {
 366                                 return array( 'uploadscripted' );
 367                         }
 368                 }
 369
 370                 /**
 371                 * Scan the uploaded file for viruses
 372                 */
 373                 $virus = $this->detectVirus( $this->mTempPath );
 374                 if ( $virus ) {
 375                         return array( 'uploadvirus', $virus );
 376                 }
 377                 wfDebug( __METHOD__ . ": all clear; passing.\n" );
 378                 return true;
 379         }
 380
 381         /**
 382          * Check whether the user can edit, upload and create the image.
 383          *
 384          * @param $user the User object to verify the permissions against
 385          * @return mixed An array as returned by getUserPermissionsErrors or true
 386          *               in case the user has proper permissions.
 387          */
 388         public function verifyPermissions( $user ) {
 389                 /**
 390                  * If the image is protected, non-sysop users won't be able
 391                  * to modify it by uploading a new revision.
 392                  */
 393                 $nt = $this->getTitle();
 394                 if( is_null( $nt ) ) {
 395                         return true;
 396                 }
 397                 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
 398                 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
 399                 if ( $nt->exists() ) {
 400                         $permErrorsCreate = $nt->getUserPermissionsErrors( 'createpage', $user );
 401                 } else {
 402                         $permErrorsCreate = array();
 403                 }
 404                 if( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
 405                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
 406                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
 407                         return $permErrors;
 408                 }
 409                 return true;
 410         }
 411
 412         /**
 413          * Check for non fatal problems with the file
 414          *
 415          * @return Array of warnings
 416          */
 417         public function checkWarnings() {
 418                 $warnings = array();
 419
 420                 $localFile = $this->getLocalFile();
 421                 $filename = $localFile->getName();
 422                 $n = strrpos( $filename, '.' );
 423                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
 424
 425                 /**
 426                  * Check whether the resulting filename is different from the desired one,
 427                  * but ignore things like ucfirst() and spaces/underscore things
 428                  */
 429                 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName );
 430                 $comparableName = Title::capitalize( $comparableName, NS_FILE );
 431
 432                 if( $this->mDesiredDestName != $filename && $comparableName != $filename ) {
 433                         $warnings['badfilename'] = $filename;
 434                 }
 435
 436                 // Check whether the file extension is on the unwanted list
 437                 global $wgCheckFileExtensions, $wgFileExtensions;
 438                 if ( $wgCheckFileExtensions ) {
 439                         if ( !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) ) {
 440                                 $warnings['filetype-unwanted-type'] = $this->mFinalExtension;
 441                         }
 442                 }
 443
 444                 global $wgUploadSizeWarning;
 445                 if ( $wgUploadSizeWarning && ( $this->mFileSize > $wgUploadSizeWarning ) ) {
 446                         $warnings['large-file'] = $wgUploadSizeWarning;
 447                 }
 448
 449                 if ( $this->mFileSize == 0 ) {
 450                         $warnings['emptyfile'] = true;
 451                 }
 452
 453                 $exists = self::getExistsWarning( $localFile );
 454                 if( $exists !== false ) {
 455                         $warnings['exists'] = $exists;
 456                 }
 457
 458                 // Check dupes against existing files
 459                 $hash = File::sha1Base36( $this->mTempPath );
 460                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 461                 $title = $this->getTitle();
 462                 // Remove all matches against self
 463                 foreach ( $dupes as $key => $dupe ) {
 464                         if( $title->equals( $dupe->getTitle() ) ) {
 465                                 unset( $dupes[$key] );
 466                         }
 467                 }
 468                 if( $dupes ) {
 469                         $warnings['duplicate'] = $dupes;
 470                 }
 471
 472                 // Check dupes against archives
 473                 $archivedImage = new ArchivedFile( null, 0, "{$hash}.{$this->mFinalExtension}" );
 474                 if ( $archivedImage->getID() > 0 ) {
 475                         $warnings['duplicate-archive'] = $archivedImage->getName();
 476                 }
 477
 478                 return $warnings;
 479         }
 480
 481         /**
 482          * Really perform the upload. Stores the file in the local repo, watches
 483          * if necessary and runs the UploadComplete hook.
 484          *
 485          * @return mixed Status indicating the whether the upload succeeded.
 486          */
 487         public function performUpload( $comment, $pageText, $watch, $user ) {
 488                 wfDebug( "\n\n\performUpload: sum: " . $comment . ' c: ' . $pageText .
 489                         ' w: ' . $watch );
 490                 $status = $this->getLocalFile()->upload( $this->mTempPath, $comment, $pageText,
 491                         File::DELETE_SOURCE, $this->mFileProps, false, $user );
 492
 493                 if( $status->isGood() ) {
 494                         if ( $watch ) {
 495                                 $user->addWatch( $this->getLocalFile()->getTitle() );
 496                         }
 497
 498                         wfRunHooks( 'UploadComplete', array( &$this ) );
 499                 }
 500
 501                 return $status;
 502         }
 503
 504         /**
 505          * Returns the title of the file to be uploaded. Sets mTitleError in case
 506          * the name was illegal.
 507          *
 508          * @return Title The title of the file or null in case the name was illegal
 509          */
 510         public function getTitle() {
 511                 if ( $this->mTitle !== false ) {
 512                         return $this->mTitle;
 513                 }
 514
 515                 /**
 516                  * Chop off any directories in the given filename. Then
 517                  * filter out illegal characters, and try to make a legible name
 518                  * out of it. We'll strip some silently that Title would die on.
 519                  */
 520                 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mDesiredDestName );
 521                 /* Normalize to title form before we do any further processing */
 522                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 523                 if( is_null( $nt ) ) {
 524                         $this->mTitleError = self::ILLEGAL_FILENAME;
 525                         return $this->mTitle = null;
 526                 }
 527                 $this->mFilteredName = $nt->getDBkey();
 528
 529                 /**
 530                  * We'll want to blacklist against *any* 'extension', and use
 531                  * only the final one for the whitelist.
 532                  */
 533                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 534
 535                 if( count( $ext ) ) {
 536                         $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
 537                 } else {
 538                         $this->mFinalExtension = '';
 539                 }
 540
 541                 /* Don't allow users to override the blacklist (check file extension) */
 542                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
 543                 global $wgFileExtensions, $wgFileBlacklist;
 544                 if ( $this->mFinalExtension == '' ) {
 545                         $this->mTitleError = self::FILETYPE_MISSING;
 546                         return $this->mTitle = null;
 547                 } elseif ( $this->checkFileExtensionList( $ext, $wgFileBlacklist ) ||
 548                                 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
 549                                         !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) ) ) {
 550                         $this->mTitleError = self::FILETYPE_BADTYPE;
 551                         return $this->mTitle = null;
 552                 }
 553
 554                 # If there was more than one "extension", reassemble the base
 555                 # filename to prevent bogus complaints about length
 556                 if( count( $ext ) > 1 ) {
 557                         for( $i = 0; $i < count( $ext ) - 1; $i++ ) {
 558                                 $partname .= '.' . $ext[$i];
 559                         }
 560                 }
 561
 562                 if( strlen( $partname ) < 1 ) {
 563                         $this->mTitleError =  self::MIN_LENGTH_PARTNAME;
 564                         return $this->mTitle = null;
 565                 }
 566
 567                 return $this->mTitle = $nt;
 568         }
 569
 570         /**
 571          * Return the local file and initializes if necessary.
 572          */
 573         public function getLocalFile() {
 574                 if( is_null( $this->mLocalFile ) ) {
 575                         $nt = $this->getTitle();
 576                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
 577                 }
 578                 return $this->mLocalFile;
 579         }
 580
 581         /**
 582          * Stash a file in a temporary directory for later processing
 583          * after the user has confirmed it.
 584          *
 585          * If the user doesn't explicitly cancel or accept, these files
 586          * can accumulate in the temp directory.
 587          *
 588          * @param $saveName String: the destination filename
 589          * @param $tempSrc String: the source temporary file to save
 590          * @return String: full path the stashed file, or false on failure
 591          */
 592         protected function saveTempUploadedFile( $saveName, $tempSrc ) {
 593                 $repo = RepoGroup::singleton()->getLocalRepo();
 594                 $status = $repo->storeTemp( $saveName, $tempSrc );
 595                 return $status;
 596         }
 597
 598         /**
 599          * Stash a file in a temporary directory for later processing,
 600          * and save the necessary descriptive info into the session.
 601          * Returns a key value which will be passed through a form
 602          * to pick up the path info on a later invocation.
 603          *
 604          * @return Integer: session key
 605          */
 606         public function stashSession() {
 607                 $status = $this->saveTempUploadedFile( $this->mDestName, $this->mTempPath );
 608                 if( !$status->isOK() ) {
 609                         # Couldn't save the file.
 610                         return false;
 611                 }
 612
 613                 $key = $this->getSessionKey();
 614                 $_SESSION[self::SESSION_KEYNAME][$key] = array(
 615                         'mTempPath'       => $status->value,
 616                         'mFileSize'       => $this->mFileSize,
 617                         'mFileProps'      => $this->mFileProps,
 618                         'version'         => self::SESSION_VERSION,
 619                 );
 620                 return $key;
 621         }
 622
 623         /**
 624          * Generate a random session key from stash in cases where we want
 625          * to start an upload without much information
 626          */
 627         protected function getSessionKey() {
 628                 $key = mt_rand( 0, 0x7fffffff );
 629                 $_SESSION[self::SESSION_KEYNAME][$key] = array();
 630                 return $key;
 631         }
 632
 633         /**
 634          * If we've modified the upload file we need to manually remove it
 635          * on exit to clean up.
 636          */
 637         public function cleanupTempFile() {
 638                 if ( $this->mRemoveTempFile && $this->mTempPath && file_exists( $this->mTempPath ) ) {
 639                         wfDebug( __METHOD__ . ": Removing temporary file {$this->mTempPath}\n" );
 640                         unlink( $this->mTempPath );
 641                 }
 642         }
 643
 644         public function getTempPath() {
 645                 return $this->mTempPath;
 646         }
 647
 648         /**
 649          * Split a file into a base name and all dot-delimited 'extensions'
 650          * on the end. Some web server configurations will fall back to
 651          * earlier pseudo-'extensions' to determine type and execute
 652          * scripts, so the blacklist needs to check them all.
 653          *
 654          * @return array
 655          */
 656         public static function splitExtensions( $filename ) {
 657                 $bits = explode( '.', $filename );
 658                 $basename = array_shift( $bits );
 659                 return array( $basename, $bits );
 660         }
 661
 662         /**
 663          * Perform case-insensitive match against a list of file extensions.
 664          * Returns true if the extension is in the list.
 665          *
 666          * @param $ext String
 667          * @param $list Array
 668          * @return Boolean
 669          */
 670         public static function checkFileExtension( $ext, $list ) {
 671                 return in_array( strtolower( $ext ), $list );
 672         }
 673
 674         /**
 675          * Perform case-insensitive match against a list of file extensions.
 676          * Returns true if any of the extensions are in the list.
 677          *
 678          * @param $ext Array
 679          * @param $list Array
 680          * @return Boolean
 681          */
 682         public static function checkFileExtensionList( $ext, $list ) {
 683                 foreach( $ext as $e ) {
 684                         if( in_array( strtolower( $e ), $list ) ) {
 685                                 return true;
 686                         }
 687                 }
 688                 return false;
 689         }
 690
 691         /**
 692          * Checks if the mime type of the uploaded file matches the file extension.
 693          *
 694          * @param $mime String: the mime type of the uploaded file
 695          * @param $extension String: the filename extension that the file is to be served with
 696          * @return Boolean
 697          */
 698         public static function verifyExtension( $mime, $extension ) {
 699                 $magic = MimeMagic::singleton();
 700
 701                 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' )
 702                         if ( !$magic->isRecognizableExtension( $extension ) ) {
 703                                 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
 704                                         "unrecognized extension '$extension', can't verify\n" );
 705                                 return true;
 706                         } else {
 707                                 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; ".
 708                                         "recognized extension '$extension', so probably invalid file\n" );
 709                                 return false;
 710                         }
 711
 712                 $match = $magic->isMatchingExtension( $extension, $mime );
 713
 714                 if ( $match === null ) {
 715                         wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
 716                         return true;
 717                 } elseif( $match === true ) {
 718                         wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
 719
 720                         #TODO: if it's a bitmap, make sure PHP or ImageMagic resp. can handle it!
 721                         return true;
 722
 723                 } else {
 724                         wfDebug( __METHOD__ . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
 725                         return false;
 726                 }
 727         }
 728
 729         /**
 730          * Heuristic for detecting files that *could* contain JavaScript instructions or
 731          * things that may look like HTML to a browser and are thus
 732          * potentially harmful. The present implementation will produce false
 733          * positives in some situations.
 734          *
 735          * @param $file String: pathname to the temporary upload file
 736          * @param $mime String: the mime type of the file
 737          * @param $extension String: the extension of the file
 738          * @return Boolean: true if the file contains something looking like embedded scripts
 739          */
 740         public static function detectScript( $file, $mime, $extension ) {
 741                 global $wgAllowTitlesInSVG;
 742
 743                 # ugly hack: for text files, always look at the entire file.
 744                 # For binary field, just check the first K.
 745
 746                 if( strpos( $mime,'text/' ) === 0 ) {
 747                         $chunk = file_get_contents( $file );
 748                 } else {
 749                         $fp = fopen( $file, 'rb' );
 750                         $chunk = fread( $fp, 1024 );
 751                         fclose( $fp );
 752                 }
 753
 754                 $chunk = strtolower( $chunk );
 755
 756                 if( !$chunk ) {
 757                         return false;
 758                 }
 759
 760                 # decode from UTF-16 if needed (could be used for obfuscation).
 761                 if( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
 762                         $enc = 'UTF-16BE';
 763                 } elseif( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
 764                         $enc = 'UTF-16LE';
 765                 } else {
 766                         $enc = null;
 767                 }
 768
 769                 if( $enc ) {
 770                         $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
 771                 }
 772
 773                 $chunk = trim( $chunk );
 774
 775                 # FIXME: convert from UTF-16 if necessarry!
 776                 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
 777
 778                 # check for HTML doctype
 779                 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
 780                         return true;
 781                 }
 782
 783                 /**
 784                  * Internet Explorer for Windows performs some really stupid file type
 785                  * autodetection which can cause it to interpret valid image files as HTML
 786                  * and potentially execute JavaScript, creating a cross-site scripting
 787                  * attack vectors.
 788                  *
 789                  * Apple's Safari browser also performs some unsafe file type autodetection
 790                  * which can cause legitimate files to be interpreted as HTML if the
 791                  * web server is not correctly configured to send the right content-type
 792                  * (or if you're really uploading plain text and octet streams!)
 793                  *
 794                  * Returns true if IE is likely to mistake the given file for HTML.
 795                  * Also returns true if Safari would mistake the given file for HTML
 796                  * when served with a generic content-type.
 797                  */
 798                 $tags = array(
 799                         '<a href',
 800                         '<body',
 801                         '<head',
 802                         '<html',   #also in safari
 803                         '<img',
 804                         '<pre',
 805                         '<script', #also in safari
 806                         '<table'
 807                 );
 808
 809                 if( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
 810                         $tags[] = '<title';
 811                 }
 812
 813                 foreach( $tags as $tag ) {
 814                         if( false !== strpos( $chunk, $tag ) ) {
 815                                 return true;
 816                         }
 817                 }
 818
 819                 /*
 820                  * look for JavaScript
 821                  */
 822
 823                 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
 824                 $chunk = Sanitizer::decodeCharReferences( $chunk );
 825
 826                 # look for script-types
 827                 if( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
 828                         return true;
 829                 }
 830
 831                 # look for html-style script-urls
 832                 if( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
 833                         return true;
 834                 }
 835
 836                 # look for css-style script-urls
 837                 if( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
 838                         return true;
 839                 }
 840
 841                 wfDebug( __METHOD__ . ": no scripts found\n" );
 842                 return false;
 843         }
 844
 845         protected function detectScriptInSvg( $filename ) {
 846                 $check = new XmlTypeCheck( $filename, array( $this, 'checkSvgScriptCallback' ) );
 847                 return $check->filterMatch;
 848         }
 849
 850         /**
 851          * @todo Replace this with a whitelist filter!
 852          */
 853         public function checkSvgScriptCallback( $element, $attribs ) {
 854                 $stripped = $this->stripXmlNamespace( $element );
 855
 856                 if( $stripped == 'script' ) {
 857                         wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
 858                         return true;
 859                 }
 860
 861                 foreach( $attribs as $attrib => $value ) {
 862                         $stripped = $this->stripXmlNamespace( $attrib );
 863                         if( substr( $stripped, 0, 2 ) == 'on' ) {
 864                                 wfDebug( __METHOD__ . ": Found script attribute '$attrib'='value' in uploaded file.\n" );
 865                                 return true;
 866                         }
 867                         if( $stripped == 'href' && strpos( strtolower( $value ), 'javascript:' ) !== false ) {
 868                                 wfDebug( __METHOD__ . ": Found script href attribute '$attrib'='$value' in uploaded file.\n" );
 869                                 return true;
 870                         }
 871                 }
 872         }
 873
 874         private function stripXmlNamespace( $name ) {
 875                 // 'http://www.w3.org/2000/svg:script' -> 'script'
 876                 $parts = explode( ':', strtolower( $name ) );
 877                 return array_pop( $parts );
 878         }
 879
 880         /**
 881          * Generic wrapper function for a virus scanner program.
 882          * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
 883          * $wgAntivirusRequired may be used to deny upload if the scan fails.
 884          *
 885          * @param $file String: pathname to the temporary upload file
 886          * @return mixed false if not virus is found, NULL if the scan fails or is disabled,
 887          *         or a string containing feedback from the virus scanner if a virus was found.
 888          *         If textual feedback is missing but a virus was found, this function returns true.
 889          */
 890         public static function detectVirus( $file ) {
 891                 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
 892
 893                 if ( !$wgAntivirus ) {
 894                         wfDebug( __METHOD__ . ": virus scanner disabled\n" );
 895                         return null;
 896                 }
 897
 898                 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
 899                         wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
 900                         $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
 901                                 array( 'virus-badscanner', $wgAntivirus ) );
 902                         return wfMsg( 'virus-unknownscanner' ) . " $wgAntivirus";
 903                 }
 904
 905                 # look up scanner configuration
 906                 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
 907                 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
 908                 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
 909                         $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
 910
 911                 if ( strpos( $command, "%f" ) === false ) {
 912                         # simple pattern: append file to scan
 913                         $command .= " " . wfEscapeShellArg( $file );
 914                 } else {
 915                         # complex pattern: replace "%f" with file to scan
 916                         $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
 917                 }
 918
 919                 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
 920
 921                 # execute virus scanner
 922                 $exitCode = false;
 923
 924                 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
 925                 #      that does not seem to be worth the pain.
 926                 #      Ask me (Duesentrieb) about it if it's ever needed.
 927                 $output = wfShellExec( "$command 2>&1", $exitCode );
 928
 929                 # map exit code to AV_xxx constants.
 930                 $mappedCode = $exitCode;
 931                 if ( $exitCodeMap ) {
 932                         if ( isset( $exitCodeMap[$exitCode] ) ) {
 933                                 $mappedCode = $exitCodeMap[$exitCode];
 934                         } elseif ( isset( $exitCodeMap["*"] ) ) {
 935                                 $mappedCode = $exitCodeMap["*"];
 936                         }
 937                 }
 938
 939                 if ( $mappedCode === AV_SCAN_FAILED ) {
 940                         # scan failed (code was mapped to false by $exitCodeMap)
 941                         wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
 942
 943                         if ( $wgAntivirusRequired ) {
 944                                 return wfMsg( 'virus-scanfailed', array( $exitCode ) );
 945                         } else {
 946                                 return null;
 947                         }
 948                 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
 949                         # scan failed because filetype is unknown (probably imune)
 950                         wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
 951                         return null;
 952                 } elseif ( $mappedCode === AV_NO_VIRUS ) {
 953                         # no virus found
 954                         wfDebug( __METHOD__ . ": file passed virus scan.\n" );
 955                         return false;
 956                 } else {
 957                         $output = trim( $output );
 958
 959                         if ( !$output ) {
 960                                 $output = true; #if there's no output, return true
 961                         } elseif ( $msgPattern ) {
 962                                 $groups = array();
 963                                 if ( preg_match( $msgPattern, $output, $groups ) ) {
 964                                         if ( $groups[1] ) {
 965                                                 $output = $groups[1];
 966                                         }
 967                                 }
 968                         }
 969
 970                         wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
 971                         return $output;
 972                 }
 973         }
 974
 975         /**
 976          * Check if the temporary file is MacBinary-encoded, as some uploads
 977          * from Internet Explorer on Mac OS Classic and Mac OS X will be.
 978          * If so, the data fork will be extracted to a second temporary file,
 979          * which will then be checked for validity and either kept or discarded.
 980          */
 981         private function checkMacBinary() {
 982                 $macbin = new MacBinary( $this->mTempPath );
 983                 if( $macbin->isValid() ) {
 984                         $dataFile = tempnam( wfTempDir(), 'WikiMacBinary' );
 985                         $dataHandle = fopen( $dataFile, 'wb' );
 986
 987                         wfDebug( __METHOD__ . ": Extracting MacBinary data fork to $dataFile\n" );
 988                         $macbin->extractData( $dataHandle );
 989
 990                         $this->mTempPath = $dataFile;
 991                         $this->mFileSize = $macbin->dataForkLength();
 992
 993                         // We'll have to manually remove the new file if it's not kept.
 994                         $this->mRemoveTempFile = true;
 995                 }
 996                 $macbin->close();
 997         }
 998
 999         /**
1000          * Check if there's an overwrite conflict and, if so, if restrictions
1001          * forbid this user from performing the upload.
1002          *
1003          * @return mixed true on success, error string on failure
1004          */
1005         private function checkOverwrite() {
1006                 global $wgUser;
1007                 // First check whether the local file can be overwritten
1008                 $file = $this->getLocalFile();
1009                 if( $file->exists() ) {
1010                         if( !self::userCanReUpload( $wgUser, $file ) ) {
1011                                 return 'fileexists-forbidden';
1012                         } else {
1013                                 return true;
1014                         }
1015                 }
1016
1017                 /* Check shared conflicts: if the local file does not exist, but
1018                  * wfFindFile finds a file, it exists in a shared repository.
1019                  */
1020                 $file = wfFindFile( $this->getTitle() );
1021                 if ( $file && !$wgUser->isAllowed( 'reupload-shared' ) ) {
1022                         return 'fileexists-shared-forbidden';
1023                 }
1024
1025                 return true;
1026         }
1027
1028         /**
1029          * Check if a user is the last uploader
1030          *
1031          * @param $user User object
1032          * @param $img String: image name
1033          * @return Boolean
1034          */
1035         public static function userCanReUpload( User $user, $img ) {
1036                 if( $user->isAllowed( 'reupload' ) ) {
1037                         return true; // non-conditional
1038                 }
1039                 if( !$user->isAllowed( 'reupload-own' ) ) {
1040                         return false;
1041                 }
1042                 if( is_string( $img ) ) {
1043                         $img = wfLocalFile( $img );
1044                 }
1045                 if ( !( $img instanceof LocalFile ) ) {
1046                         return false;
1047                 }
1048
1049                 return $user->getId() == $img->getUser( 'id' );
1050         }
1051
1052         /**
1053          * Helper function that does various existence checks for a file.
1054          * The following checks are performed:
1055          * - The file exists
1056          * - Article with the same name as the file exists
1057          * - File exists with normalized extension
1058          * - The file looks like a thumbnail and the original exists
1059          *
1060          * @param $file The File object to check
1061          * @return mixed False if the file does not exists, else an array
1062          */
1063         public static function getExistsWarning( $file ) {
1064                 if( $file->exists() ) {
1065                         return array( 'warning' => 'exists', 'file' => $file );
1066                 }
1067
1068                 if( $file->getTitle()->getArticleID() ) {
1069                         return array( 'warning' => 'page-exists', 'file' => $file );
1070                 }
1071
1072                 if ( $file->wasDeleted() && !$file->exists() ) {
1073                         return array( 'warning' => 'was-deleted', 'file' => $file );
1074                 }
1075
1076                 if( strpos( $file->getName(), '.' ) == false ) {
1077                         $partname = $file->getName();
1078                         $extension = '';
1079                 } else {
1080                         $n = strrpos( $file->getName(), '.' );
1081                         $extension = substr( $file->getName(), $n + 1 );
1082                         $partname = substr( $file->getName(), 0, $n );
1083                 }
1084                 $normalizedExtension = File::normalizeExtension( $extension );
1085
1086                 if ( $normalizedExtension != $extension ) {
1087                         // We're not using the normalized form of the extension.
1088                         // Normal form is lowercase, using most common of alternate
1089                         // extensions (eg 'jpg' rather than 'JPEG').
1090                         //
1091                         // Check for another file using the normalized form...
1092                         $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
1093                         $file_lc = wfLocalFile( $nt_lc );
1094
1095                         if( $file_lc->exists() ) {
1096                                 return array(
1097                                         'warning' => 'exists-normalized',
1098                                         'file' => $file,
1099                                         'normalizedFile' => $file_lc
1100                                 );
1101                         }
1102                 }
1103
1104                 if ( self::isThumbName( $file->getName() ) ) {
1105                         # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1106                         $nt_thb = Title::newFromText( substr( $partname , strpos( $partname , '-' ) +1 ) . '.' . $extension, NS_FILE );
1107                         $file_thb = wfLocalFile( $nt_thb );
1108                         if( $file_thb->exists() ) {
1109                                 return array(
1110                                         'warning' => 'thumb',
1111                                         'file' => $file,
1112                                         'thumbFile' => $file_thb
1113                                 );
1114                         } else {
1115                                 // File does not exist, but we just don't like the name
1116                                 return array(
1117                                         'warning' => 'thumb-name',
1118                                         'file' => $file,
1119                                         'thumbFile' => $file_thb
1120                                 );
1121                         }
1122                 }
1123
1124
1125                 foreach( self::getFilenamePrefixBlacklist() as $prefix ) {
1126                         if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1127                                 return array(
1128                                         'warning' => 'bad-prefix',
1129                                         'file' => $file,
1130                                         'prefix' => $prefix
1131                                 );
1132                         }
1133                 }
1134
1135                 return false;
1136         }
1137
1138         /**
1139          * Helper function that checks whether the filename looks like a thumbnail
1140          */
1141         public static function isThumbName( $filename ) {
1142                 $n = strrpos( $filename, '.' );
1143                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
1144                 return (
1145                                         substr( $partname , 3, 3 ) == 'px-' ||
1146                                         substr( $partname , 2, 3 ) == 'px-'
1147                                 ) &&
1148                                 preg_match( "/[0-9]{2}/" , substr( $partname , 0, 2 ) );
1149         }
1150
1151         /**
1152          * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
1153          *
1154          * @return array list of prefixes
1155          */
1156         public static function getFilenamePrefixBlacklist() {
1157                 $blacklist = array();
1158                 $message = wfMsgForContent( 'filename-prefix-blacklist' );
1159                 if( $message && !( wfEmptyMsg( 'filename-prefix-blacklist', $message ) || $message == '-' ) ) {
1160                         $lines = explode( "\n", $message );
1161                         foreach( $lines as $line ) {
1162                                 // Remove comment lines
1163                                 $comment = substr( trim( $line ), 0, 1 );
1164                                 if ( $comment == '#' || $comment == '' ) {
1165                                         continue;
1166                                 }
1167                                 // Remove additional comments after a prefix
1168                                 $comment = strpos( $line, '#' );
1169                                 if ( $comment > 0 ) {
1170                                         $line = substr( $line, 0, $comment-1 );
1171                                 }
1172                                 $blacklist[] = trim( $line );
1173                         }
1174                 }
1175                 return $blacklist;
1176         }
1177
1178         public function getImageInfo( $result ) {
1179                 $file = $this->getLocalFile();
1180                 $imParam = ApiQueryImageInfo::getPropertyNames();
1181                 return ApiQueryImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1182         }
1183
1184         public function convertVerifyErrorToStatus( $error ) {
1185                 $args = func_get_args();
1186                 array_shift($args);
1187                 return Status::newFatal( $this->getVerificationErrorCode( $error ), $args );
1188         }
1189 }